Using an agent-based model to evaluate the effect of producer specialization on the epidemiological resilience of livestock production networks
Authored by Serge W Wiltshire
Date Published: 2018
DOI: 10.1371/journal.pone.0194013
Sponsors:
United States Department of Agriculture (USDA)
Platforms:
Java
AnyLogic
Model Documentation:
ODD
Flow charts
Pseudocode
Model Code URLs:
Model code not found
Abstract
An agent-based computer model that builds representative regional U.S.
hog production networks was developed and employed to assess the
potential impact of the ongoing trend towards increased producer
specialization upon network-level resilience to catastrophic disease
outbreaks. Empirical analyses suggest that the spatial distribution and
connectivity patterns of contact networks often predict epidemic
spreading dynamics. Our model heuristically generates realistic systems
composed of hog producer, feed mill, and slaughter plant agents. Network
edges are added during each run as agents exchange livestock and feed.
The heuristics governing agents' contact patterns account for factors
including their industry roles, physical proximities, and the age of
their livestock. In each run, an infection is introduced, and may spread
according to probabilities associated with the various modes of contact.
For each of three treatments-defined by one-phase, two-phase, and
three-phase production systems-a parameter variation experiment examines
the impact of the spatial density of producer agents in the system upon
the length and size of disease outbreaks. Resulting data show phase
transitions whereby, above some density threshold, systemic outbreaks
become possible, echoing findings from percolation theory. Data analysis
reveals that multi-phase production systems are vulnerable to
catastrophic outbreaks at lower spatial densities, have more abrupt
percolation transitions, and are characterized by less-predictable
outbreak scales and durations. Key differences in network-level metrics
shed light on these results, suggesting that the absence of
potentially-bridging producer-producer edges may be largely responsible
for the superior disease resilience of single-phase ``farrow to
finish{''} production systems.
Tags
Complex networks
Social networks
epidemics
Dynamics
transportation
percolation
robustness
Outbreaks
Cellular-automata
Tolerance
Mathematics
Diarrhea virus-infection
Potential disease spread
Small-world
networks
Trade patterns
Heterogeneous networks